INTERNATIONAL ,IOURNAI,^LEPROS1' VolujiLe 61, Number 4 Printed in 111(' (LS.A. Case-Control Study of BCG Vaccination as a Risk Factor for Leprosy and Tuberculosis in Western Kenyal Patrick A. Orege, Paul E. M. Fine, Sebastian B. Lucas, Michael Obura, Cornelius Okelo, and Peter Okuku2 The effect of the bacillus Calmette-Gue- rin (BCG) vaccine on mycobacterial dis- cases is still controversial despite the fact that it is one of the most widely used vac- eines (4). Major field trials in Africa and elsewhere have failed to produce consistent results on the protection by BCG against leprosy and tuberculosis (14)• A controlled trial undertaken in Uganda indicated 80% protection by BCG against tuberculoid lep- rosy (20). In Malawi (14), BCG has becn found to impart 49% protection against pauciba- cillary lcprosy and 84% protection against multibacillary leprosy. A household contact study in Togo (22) and a case-control study undertaken in Cameroon (2) both detected approximately 60% protection against tu- berculosis. The study undertakcn in Malawi failed to detect any protection of BCG against tuberculosis (")• Studies from other parts of the world have produced varying results, ranging from 20% to 60% against leprosy and O to 80% against tuberculosis (4, 1)),17,. ) • Only in South India (23) and Malawi (14) has the effectiveness of BCG becn stud- ied simultancously against tuberculosis and leprosy in the same population. In both populations higher protection was found against leprosy than against tuberculosis. ' Received for publication on 1 October 1992; ac- cepted for publication in revised form on 24 August 1993. = P. A. Orege, M.B.Ch.B., Ph.D.; M. Obura, Dip.Clin.Nled.; C. Okelo, Dip.Clin.Med.; P. Okuku, Cert. Community Nurse, Kenya Medical Research In- stitute, Alupe Lcprosy and Skin Diseases Research Center, P.O. Box 3, Busia, Kenya. P. E. M. Fine, V.M.D., Ph.D., Department of Epidemiology and Pop- ulation Sciences, London School of Hygiene and Trop- ical Medicine, Keppel Street, London WC I E 7HT, U.K. S. B. Lucas, F.R.C.P., F.R.C. Path., University College and Middlesex School of Medicine, Gower Street, Lon- don WC I E 6JJ, U.K. Despite all of the controversies and the difficulties of demonstrating the impact of BCG at a population levei, BCG is still giv- en routinely to millions of children in de- veloping countries, with the belief that it is preventing large numbers of cases of tuber- culosis (or leprosy). It is important to assess the actual effect of this program on disease incidence. In this paper wc report the fmdings of a case-control study designed to measure the protective efficacy of BCG against tuber- culosis and leprosy in Western Kenya. MATERIALS AND METHODS The study was undertaken in four dis- tricts of wcstern Kenya known to be endem- ic for both leprosy and tuberculosis: Busia, Siaya, Kisumu and South Nyan7a 12). The estimated population in the study arca in 1989 was 3.3 million (3), of whom 54% were younger than 15 years of age and 51% were females. Leprosy and tuberculosis control is coordinated in the study arca by the Na- tional Leprosy and Tuberculosis Control Program (NLTP), an integratcd control pro- gram run by the Kenya Ministry of Health. The programs are based on passive case finding and treatment. BCG was introduced into the study arca in 1962 with mass vaccination campaigns concentrating on schools and market places. The official target population was children under 16 years of age (21, 24 ) • After the initial phase, which lasted almost 10 years, the BCG program was handed over to maternal and child health (MCH) welfare programs in 1972, and newborns and children less than 5 years old continued to receive BCG from MCH. In 1984, BCG was taken over by the Kenya Expanded Program on Im- munization (KEPI), which now administers BCG vaccine to newborns and to children less than 5 years old (8). 542
Transcript
INTERNATIONAL ,IOURNAI,^LEPROS1' VolujiLe 61, Number 4Printed in 111(' (LS.A.
Case-Control Study of BCG Vaccination as a RiskFactor for Leprosy and Tuberculosis in
Western KenyalPatrick A. Orege, Paul E. M. Fine, Sebastian B. Lucas,
Michael Obura, Cornelius Okelo, and Peter Okuku2
The effect of the bacillus Calmette-Gue-rin (BCG) vaccine on mycobacterial dis-cases is still controversial despite the factthat it is one of the most widely used vac-eines (4). Major field trials in Africa andelsewhere have failed to produce consistentresults on the protection by BCG againstleprosy and tuberculosis (14)• A controlledtrial undertaken in Uganda indicated 80%protection by BCG against tuberculoid lep-rosy (20). In Malawi (14), BCG has becn foundto impart 49% protection against pauciba-cillary lcprosy and 84% protection againstmultibacillary leprosy. A household contactstudy in Togo (22) and a case-control studyundertaken in Cameroon (2) both detectedapproximately 60% protection against tu-berculosis. The study undertakcn in Malawifailed to detect any protection of BCGagainst tuberculosis (")• Studies from otherparts of the world have produced varyingresults, ranging from 20% to 60% againstleprosy and O to 80% against tuberculosis(4, 1)),17,.) • Only in South India (23) and Malawi(14) has the effectiveness of BCG becn stud-ied simultancously against tuberculosis andleprosy in the same population. In bothpopulations higher protection was foundagainst leprosy than against tuberculosis.
' Received for publication on 1 October 1992; ac-cepted for publication in revised form on 24 August1993.
= P. A. Orege, M.B.Ch.B., Ph.D.; M. Obura,Dip.Clin.Nled.; C. Okelo, Dip.Clin.Med.; P. Okuku,Cert. Community Nurse, Kenya Medical Research In-stitute, Alupe Lcprosy and Skin Diseases ResearchCenter, P.O. Box 3, Busia, Kenya. P. E. M. Fine,V.M.D., Ph.D., Department of Epidemiology and Pop-ulation Sciences, London School of Hygiene and Trop-ical Medicine, Keppel Street, London WC I E 7HT, U.K.S. B. Lucas, F.R.C.P., F.R.C. Path., University Collegeand Middlesex School of Medicine, Gower Street, Lon-don WC I E 6JJ, U.K.
Despite all of the controversies and thedifficulties of demonstrating the impact ofBCG at a population levei, BCG is still giv-en routinely to millions of children in de-veloping countries, with the belief that it ispreventing large numbers of cases of tuber-culosis (or leprosy). It is important to assessthe actual effect of this program on diseaseincidence.
In this paper wc report the fmdings of acase-control study designed to measure theprotective efficacy of BCG against tuber-culosis and leprosy in Western Kenya.
MATERIALS AND METHODSThe study was undertaken in four dis-
tricts of wcstern Kenya known to be endem-ic for both leprosy and tuberculosis: Busia,Siaya, Kisumu and South Nyan7a 12). Theestimated population in the study arca in1989 was 3.3 million (3), of whom 54% wereyounger than 15 years of age and 51% werefemales. Leprosy and tuberculosis control iscoordinated in the study arca by the Na-tional Leprosy and Tuberculosis ControlProgram (NLTP), an integratcd control pro-gram run by the Kenya Ministry of Health.The programs are based on passive casefinding and treatment.
BCG was introduced into the study arcain 1962 with mass vaccination campaignsconcentrating on schools and market places.The official target population was childrenunder 16 years of age (21, 24
) • After the initialphase, which lasted almost 10 years, theBCG program was handed over to maternaland child health (MCH) welfare programsin 1972, and newborns and children lessthan 5 years old continued to receive BCGfrom MCH. In 1984, BCG was taken overby the Kenya Expanded Program on Im-munization (KEPI), which now administersBCG vaccine to newborns and to childrenless than 5 years old (8).
542
61, 4^Orege, et al.: J3CG I"accination in Kenya^543
As far as is known, most of the BCG usedin the study arca was Glaxo (freeze-dried).The recommended dosages were 0.05 ml forchildren less than 1 ycar old and 0.1 ml forchildren older than 1 ycar (0.1 ml = ap-proximately 1.6 x 10 6 viable particles). Thevaccine was injected intradermally into thelateral aspect of the left upper forearm, justlateral to the antecubital fossa.
For the purposcs of this study, a leprosycase was defined as "a newly diagnosed, pre-viously untreated, paucibacillary or multi-bacillary case with bacteriological and/orhistopathological evidence of Ieprosy, agedbetween 0-65 years, resident in Busia, Siaya,Kisumu or South Nyanza districts, and di-agnosed/registered for treatment after 1April 1989." Neural leprosy cases were ex-cluded from the study.
A tuberculosis case was defined as "a per-son newly diagnosed clinically and/or bac-teriologically as having pulmonary tuber-culosis within the health facilities andresidents of Busia, Siaya, Kisumu and SouthNyanza districts after 1 July 1989, and agedbetween 15-44 years."
Leprosy cases self-reported to the healthunits (dispensaries, health centers or districthospitais). Potential cases were recruited intothe study at the time of registration or atfirst contact with the national control pro-gram. All of the identified cases were ex-amined by the principal investigator (PAO)or one of the field officers (medicai assistantsspecialized in tuberculosis and leprosy). Thecases were then classified clinically accord-ing to the Ridley-Jopling scale, i.e., polartubcrculoid (TT), borderline tuberculoid(BT), borderline borderline (BB), borderlinelepromatous (BL), and polar lepromatous(LL). These cases were also subcategorizedby diagnostic certainty according to pre-determined guidelines. Cases were subclas-sified clinically as "certainly" leprosy, "mostlikely" leprosy or "possibly" leprosy. Eachcase had a skin biopsy taken and sent to oneof us (SBL) for histopathological examina-tion. The interpretation of the skin biopsywas done according to guidelines developedin Malawi ( 5 ).
Tuberculosis cases were self-reporting pa-tients who had initially been identified bythe control program and whose sputum waspositive by direct microscopy. Another spu-tum sample was collected and sent to the
reference laboratory in Nairobi for repeatdirect examination and culture. At the ref-erence laboratory, direct microscopy wasdone using the auramine-phenol techniqueand reported as "negative," "scantly posi-tive," "moderately positive" or "heavilypositive." The same sputa were cultured andthe results reported within 6-8 weeks as"negative," "contaminated," "scanty,""moderately positive" or "heavily posi-tive." The identification of culture isolatesas Mvcobacterimn t uberculosis also was doneat the reference laboratory.
Controls were selected by identifying andexamining inhabitants of households nearthat of the case. The principal investigator(PAO) or field officers canvassed the neigh-borhood of cases following predeterminedrules until appropriate age- and sex-matchedcontrols were identified. Once a potencialcontrol was identified, the purpose of thestudy was explained to him or her. If s/heagreed to participate, s/he was then exam-ined to rule out tuberculosis or leprosy. Anypotential control who appeared to have lep-rosy was excluded as a control for the lep-rosy study (one control was excluded on chisbasis). Any control with a history of chroniccough was excluded as a control for tuber-culosis.
All cases and controls were examined forthe presence of a BCG scar and questionedabout socioeconomic characteristics, suchas type of house, houschold items, occu-pation, land ownership, distance from thenearest health unit, crops grown by the fam-ily, and levei of education of the head of thehousehold and the respondent.
Exposure to BCG was ascertained ini-tially by the principal investigator or one ofthe field officers who were blind neither tothe study hypothesis nor to the case or con-trol status. BCG scars were later read in-derrendently by field assistants (paramedicalhealth workers) trained in the reading ofBCG scars. Evidence of a BCG scar wasrecorded as present, absent, doubtful or un-known. Field assistants were kept blind tothe study hypothesis throughout the dura-tion of the study. An effort was also madeto keep them blind to the case or controlstatus of the study subjects by invoivingthem in the collection of information onsmallpox scars, collection and examinationof urine on the spot using chemical strips
544^ International Journal of Leprosy^ 1993
TABLE 1. Distribution by age and sex ofleprosy cases and controls admitted into theBCG vaccine efficacy study.
(URISTIX, Emmes, U.K.), and collectionand parasitological examination of stools.The largest transverse diameter of any scarwas measured and recorded in millimeters.
Data analysis was conducted using Sta-tistical Analysis System (SAS) computersoftware (SAS Institute, Inc., Cary, NorthCarolina, U.S.A., 1985), and EGRET (SER-EC, Seattle, Washington, U.S.A.). Condi-tional likelihood estimates (CMLE) and their95% confidence intervals were transformedinto disease exposure odds ratios and con-fidence limits.
RESULTSOf 182 potential leprosy cases ascertained
by the routine leprosy control program, 132were enrolled into the study. Of these cases,
131 were biopsied; 42 (32%) had skin smearstaken (by the control program). Of the 132total cases, three who satisfied both clinicaland histopathological criteria were excludedfrom the study because suitable controlscould not be identified for them. This left129 leprosy cases for whom 384 controlswere identified. The major analysis was re-stricted to those who were 44 years of ageor younger since older individuals were bomtoo early to have been included in the BCGprograms. The distribution of the 69 leprosycases and 238 controls included in theseanalyses is shown in Table 1; the distribu-tion of BCG scars among the cases and con-trols is shown in Table 2.
Among the 144 eligible tuberculosis cases,13 (9%) did not have sputum sent to thereference laboratory for logistic reasons, andI 1 (7.6%) were both microscopy and culturenegative at the reference laboratory. A cer-tainty group subclassification was estab-lished to define "certain" tuberculosis (thosewho were repeat smear and/or culture pos-itive) and "possible" tuberculosis (those whowere negative to repeat smear or culture,and those who did not have sputum sent tothe central laboratory). A further 15 caseshad to be excluded since they could not befound by field assistants. Following thesecriteria, there were 119 "certain" and 25"possible" tuberculosis cases included in theanalysis (Table 3). The distribution of BCG
Males Fe- Totalmales
TABLE 2. Distribution of BCG scars by age and sex among leprosy cases and controls.a
scar status among tuberculosis cases andcontrols is shown in Table 4.
In order to take matching of controls andadditional potential confounding into ac-count, multiple conditional logistic regres-sion was performed. The estimated vaccineefficacy of BCG against leprosy is shown inTable 5; against tuberculosis in Table 6. Theoverall protective efficacy of BCG againstleprosy was found to be 81% and highlystatistically significant (95% CI = 67, 80)with some evidente (not statistically signif-icant) for higher efficacy among older in-dividuais and among females. Among tu-berculosis cases, the point estimate overallvaccine efficacy was found to be 22%, butthis was not statistically significantly greaterthan zero (95% CI = —20, 51). This esti-mated vaccine efficacy appeared to have noage or sex trend.
TABLE 5. Protectiv'e efficacy of BCG rac-cination against leprosy calculated frommatched-set case-control analvsis by age andsex.
Table 7 shows BCG vaccine efficacyagainst leprosy separated by histopatholog-ical classification and diagnostic certainty.The vaccine appeared to give similar pro-tection against ali clinicai forms of the dis-case. When cases were classified into twobroad categories of paucibacillary (TT +BT), and multibacillary leprosy (BB + BL+ LL), the protective effects were 83% (95%
TABLE 4. Distribution of BCG scar status by age and sex among tuberculosis cases andcontrols."
Age Cases Controls
(yr) +ve —ve ± U Total +ve — ve ± U Total
Males10-19 2 1 O 2 5 14 3 1 O 1820-29 19 12 O 4 35 78 28 3 O 10930-39 18 6 O 4 28 60 19 1 O 8040-49 O 6 O 3 9 11 I1 2 O 24
Total 39 25 O I 3 77 163 61 7 O 231(51%) (32%) (0%) (17%) (100%) (71%) (26%) (3%) (100%)
Females10-19 5 7 O O 12 32 7 3 O 4220-29 20 4 1 2 27 53 22 2 O 7530-39 10 11 O O 21 27 39 1 6740-49 O 7 O O 7 1 16 O O 17
Total 35 29 1 2 67 113 84 6 O 201(52%) (43%) (2%) (3%) (100%) (56%) (42%) (3%) — (100%)
CI = 85, 92) and 76% (95% CI = 30, 91),respectively, suggesting that BCG wasequally protective against paucibacillary andmultibacillary leprosy. Higher efficacy (84%)was noted against "certain" than against"most likely/possible/unknown" disease(59%).
Table 8 shows vaceine efficacy estimatesagainst tuberculosis when broken down bydiagnostic certainty. Surprisingly, althoughthe estimates were not statistically different,
TABLE 8. Protective effect of BCG vacci-'ration against tuberculosis calculated frommatched-set case-control analysis adjustingfor diagnostic certainly.
a lower protective effect was found againstthe "certain" than against the "possible"category.
DISCUSSIONThe results presented here are based on
a case-control design with self-presentingleprosy and tuberculosis cases and age-, sex-and neighborhood-matched controls. Therewas no active case detection in the studyarca, thus cases in this investigation wererepresentative of those seen within the con-trol program.
This investigation provides evidence thatBCG gives stronger protection against lep-rosy than against tuberculosis in this arcaof western Kenya. This is consistent withfindings in other arcas where BCG efficacyagainst tuberculosis and leprosy have beencvaluated concurrently. In South India (23)the protective efficacy was found to be 0%and 23% for tuberculosis and leprosy, re-
TABLE 7. Protective efficacy of BCG against leprosy calculated from matched-set case-control analysis adjusting for histopathological classification and dia gnostic certainty.
spectively. In Malawi ( 7.' 4 ) the protectionwas found to be 0% and above 50% for tu-berculosis and leprosy, respectively. Esti-mates of BCG protection obtained in thepresent study were 22% for tuberculosis and81% for leprosy.
This difference is unlikely to be due tomethodological crrors for several reasons:a) BCG scar status was assesscd by fieldassistants who were purposely kept blind ofthe case or control status, and the study hy-pothesis. This should have minimized anypotential information bias, and b) we tookprecautions in ensuring the accuracy of di-agnoses. Since it is widely recognized thatthe diagnosis of leprosy is difficult, strictcriteria were developed for the diagnosis ofleprosy in the field, and also histopatholog-ically. Cases ofchildhood tuberculosis werenot included in this study.
Another possible source of error is themisclassification of BCG scars ( 6•'1. Inas-much as BCG may not leave a scar or a scarmay be missed on examination, random ornonrandom misclassification could be in-troduced into the study. In Kenya ( 8 ) it wasfound in one KEPI survcy that 5% of thechildren who had received vaccination atbirth did not have BCG scars 20 monthslazer. An independent study to estimate thesensitivity and specificity for BCG scarreading in the context of the present inves-tigation detected that 14% of the childrenwho had documented evidence of prior BCGvaccination did not have BCG scars on ex-amination. This implies a nondifferentialmisclassification, which would have low-ered the estimated protective effect. Thesensitivity and specificity of BCG scar read-ing in this study were estimated to be 86.4%and 97.4%, respectively. Fine, et al. (') havenoted from their work in Malawi that a sen-sitivity of 90% and specificity of 90% wouldlead to about 11% underestimation ofa vac-cine's efficacy. Since vaccine efficacy de-pends mainly upon the specificity of BCGscar reading ( 15 ), the "true" efficacy mayhave been underestimated in this study byabout 5%.
The protective efficacy of BCG versusleprosy in the study population was foundto be high. The protective efficacy was foundto be higher against borderline leprosy thanagainst polar forms of leprosy, bui did notdiffer between multibacillary and pauciba-
cillary disease. Protection against the mul-tibacillary form of leprosy also has beenfound in several studies (10, ", 14.23 ) , and hassignificant public health implications be-cause multibacillary disease is thought to bea major source for the spread of RI. lepraein the community.
This study detected a nonsignificant 22%protective effect of BCG against tubercu-losis. While the finding was consistent withestimates obtained from some paris of theworld ( 13. u. ' 8 19 ), it differs from studies inWest Africa (2, 22 ) which detected 60% pro-tection.
Among the possible explanations for thelow protective efficacy of BCG against tu-berculosis obtained in this study are: a)overmatching on some socioeconomic vari-ables which were associated strongly with13CG exposure, which could have existedand were noz measured; b) the existence ofatypical mycobacteria in the study arcawhich would havc lowered the observableprotective cfficacy ( 7 ) and/or c) a prepon-derante of exogenous reinfection tubercu-losis against which BCG may be able to playonly a minor role.
We emphasize that the vaccine effectsidentified in this study reflect effects againstpassively ascertained disease, which tendsto be more advanced than disease identifiedin active case-detection surveys. Consider-ing our results by classification and diag-nostic ccrtainty (Table 7), there is evidencefor strong protection against BT disease, asnoted in New Guinea ('), but little evidencefor a general shift of the disease from mul-tibacillary to paucibacillary and indeter-minate forms as observed in South India(II).
There is some evidence from this study,and from Malawi ( 14 ), of an inverse rela-tionship betwcen BCG protection againsttuberculosis and against leprosy. Therefore,it would be prudent to conduct further case-control studies in situations similar to oursin order to determine the regional variabil-ity and other determinants of BCG's effects.
SUMMARYA case-control study was carried out in
western Kenya to measure the protectionimparted by BCG against leprosy and tu-berculosis. The study involved 69 newly di-agnosed leprosy cases, 238 age-, sex- and
548^ International Journal of Leprosy^ 1993
neighborhood-matched controls, and 144newly diagnosed, sputum-smear-positivetuberculosis cases along with 432 age-, sex-and neighborhood-matched controls. Infor-mation on BCG vaccination history was in-ferred from scars. Using matched analysis,the protection imparted by BCG againstleprosy was estimated to be 81% [95% con-fidence interval (CI) = 67-90] with no ap-parent difference in protection against pau-cibacillary [vaccine efficacy (VE) = 83%,95% CI = 58-92] and multibacillary leprosy(VE = 76%; 95% CI = 30-91). The effec-tiveness against tuberculosis was apprecia-bly lower (VE = 22%) and was not statis-tically significant (95% CI = —20-51).
RESUMENSe realizó un estudio en Kenia occidental para medir
la protección conferida por el BCG contra la lepra y iatuberculosis. El estudio involucró 69 casos de leprarecién diagnosticados y 238 controles sanos apareadosen cuanto a edad, sexo y residencia, además de 144pacientes con tuberculosis de reciente diagnóstico (conbaciloscopias positivas) y 432 controles sanos aparea-dos también por edad, sexo y residencia.
La información sobre ia vacunación previa con BCGse infirió de la presencia de cicatrices. Usando análisisapareados se calculó que la protección conferida porcl BCG contra la lepra fue del 81% [intervalo de con-fianza ai 95% (CI) = 67-90], sin diferencias aparentesen ia protección contra la lepra paucibacilar [eficienciade ia vacuna (VE) = 83%; Cl ai 95% = 58-92] y lalepra multibacilar (VE = 76%; Cl al 95% = 30-91).La efectividad contra la tuberculosis fue apreciable-mente menor (VE = 22%) y no fue estadisticamentesignificativa (CI ai 95% = —20-51).
RÉSUMÉUne étude cas-témoins a été réalisée dans l'Ouest du
Kenya dans le but de mesurer la protection apportéepar le BCG vis-à-vis de la lépre et de ia tuberculose.L'étucle comprenait 69 cas de lèpre nouvellement diag-nostiqués, 238 témoins appariés pour l'âge, le sexe etle voisinage, ainsi que 144 cas de tuberculose nouvel-lement diagnostiques, positifs à l'examen direct et 432témoins appariés pour l'âge, le sexe et le voisinage. Lesantécedents de vaccination par le BCG ont été déduitsde la presence ou non de cicatrices. Sur base de testspairés, on a estime la protection apportée par le BCGvis-à-vis de la lepre à 81 % [intervalle de confiance à95 % (IC): 67-90], sans difrerence apparente dans laprotection vis-à-vis de ia lèpre paucibacillaire [effica-cité vaccinale (EV) = 83%; IC à 95% = 58-92] et lalèpre multibacillaire (EV = 76%; IC à 95% = 30-91].L'efficacité vis-à-vis de la tuberculose était notable-ment plus basse (EV = 22%) et non statistiquementsignificative (IC à 95% = —20-51).
Acknovs ledgment. This study was supported by agrant from the Special Programme for Research andTraining in Tropical Diseases (TDR) of the UNDP/World Bank/World Health Organization. We are grate-ful to GPA/WHO for partially funding the field work.The authors are grateful to Drs. J. Omwega, formerlyDirector, Respiratory Diseases Research Center, andB. Swai for their assistance. We are also thankful toDr. Jonathan Sterne for help in data analysis. Last butnot least, the authors are grateful to the Director, KenyaMedical Research Institute (KEMRI), for his permis-sion to publish this article.
REFERENCES1. BAGSHAWE, A., SCOTT, G. C., RUSSELL, D. A.,
WIGLEY, S. C., MERIANOS, A. and BERRY, G. BCGvaccination in leprosy: final results of the trial inKarimui, Papua New Guinea. Bull. WHO 67 (1989)389-399.
2. BLIN, P., DELOLME, H. G., HEYRAUD, J. D., CHAR-PAK, Y. and SENTILHES, L. Evaluation of protec-tive effect of BCG vaccination by case control studyin Yaounde, Cameroon. Tubercle 67 (1986) 283—288.
3. CENTRAL BUREAU OF STATISTICS Population Pro-jections for Kenya 1983-2000. Kenya NationalCensus 1979 Ministry of Planning and NationalDevelopment, 1983.
4. FINE, P. E. M. BCG vaccination against tuber-culosis and leprosy. Br. Med. Bull. 44 (1988) 691—703.
5. FINE, P. E. M., JOB, K., McDouGALL, A. C.,
MEYERS, W. M. and PONNIGHAUS, J. M. COM-parability among histopathologists in the diag-nosis and classification of lesions suspected of lep-rosy in Malawi. Int. J. Lepr. 54 (1986) 614-625.
6. FINE, P. E. M., PõNNIGHAUS, J. M. and MAINE, N.The distribution and implication of BCG scars inNorthern Malawi. Bull. WHO 67 (1989) 35-41.
7. FINE, P. E. M., PõNNIGHAUS, J. M., MAINE, N.,CLARKSON, J. A. and Buss, L. Protective eflicacyof BCG against leprosy in northern Malawi. Lan-cet 2 (1986) 499-502.
8. KENYA EXPANDED PROGRAMME ON IMMUNIZATION
MANAGEMENT UNIT. National ImmunizationCoverage Survey (1987). Ministry of Health, Ke-nya.
9. LEIKER, D. L., OTSYULA, Y. and ZIEDSES DES
PLANTES, M. Leprosy and tuberculosis in Kenya.Lepr. Rev. 39 (1968) 79-83.
10. LwiN, K., SUNDARESAN, T., GYI, M. M., BECHELLI,
L. M., TAMONDONG, C., GARBAJOSA, P. G.,SANSARHICQ, H. AND NOORDEEN, S. K. BCG vac-cination of children against leprosy; 14 year find-ing of the Burma trial. Bull. WHO 63 (1985) 1069—1078.
11. Muliyil, J., Nelson, K. E. and DIAMOND, E. L. Theeffect of BCG on the risk of leprosy in an endemicarca: a case control study. Int. J. Lepr. 59 (1991)"”9—'36.
61, 4^Orege, et al.: BCG Vaccination in Kenya^549
12. NATIONAL LEPROSY AND TUnERCULOSIS CONTROLPROGRAMME (NLTR). Western and NyanzaProvinces, Annual Report (1986). Ministry ofHealth, Kenya.
13. PATEL, A., SCHOFIELD, F., SISKIND, V., AIIRAHAMS,E. and PARKER, J. Case control evaluation ofschoolage BCG vaccination programme in subtropicalAustralia. Bull. WHO 69 (1991) 425-433.
14. PõNNIGIIAUS, J. M., FINE, P. E. M., STERNE, J. A.C., WILSON, R. S., MSOSA E., GRUER, P. J. K.,JENKINS, P. A., LUCAS, S. B., LIOMBA, G. and BLISS,L. Efficacy of BCG against leprosy and tubercu-losis in northern Malawi. Lancet 339 (1992) 636-639.
15. RADHAKRISIINA, S., NAIR, N. G. K. and JAYAIIAL,P. Implications of misdiagnosis in field trials ofvaccines. Indian J. Med. Res. 80 (1984) 711-720.
16. Shapiro, C., Cook, N., Evans, D., Willet, W., Fa-jarda, 1., Koch-Weser, D., Bergonzoli, G., Bola-nos, O., Guerrero, R. and HENNEKENS, C. A casecontrol study of BCG and childhood tuberculosisin Cali, Colombia. Int. J. Epidemiol. 14 (1985)441-446.
17. SMITH, P. G. Case control studies of the efficacyof BCG against tuberculosis: Tuberculosis and Re-spiratory Diseases. Papers presentcd at the XXVIthWorld Conference of International Union AgainstTuberculosis, Singapore, 4-7 Nov 1986.) Profes-
sional Postgraduate Services, Japan 1987, pp. 73-79.
18. SMITH, P.G. Case control studies of the efficacyof BCG vaccine in children. Bull. Int. Union Tu-bercul. Lung Dis. 62 (1987) 70-76.
19. SMITII, P. G. Evaluating interventions againsttropical diseases. Int. J. Epidemiol. 16 (1987) 159-166.
20. STANLEY, S. J., HOWLAND, C., STONE, M. M. andSUTHERLAND, I. BCG vaccinations against lep-rosy in Uganda: final results. J. Hyg. (Cambridge)87 (1981) 233-248.
21. SWAI, O. B., ALUOCH, J. A., KAMUNVI, F., AG-WADA, R., KWAMANGA, D. and KENYA, P. R. Na-tional BCG scar survey in the estimation of BCGvaccination in Kenya. East Afr. Mcd. J. 62 (1985)842-851.
22. TIDJANI, O., AMEDONE, A. and TEN DAM, H. G.The protective effect of BCG vaccination of thenewborn against childhood tuberculosis in an Af-rican community. Tubercle 67 (1986) 269-281.
23. TRIPATHY, S. P. The case for BCG. Ann. Natl.Acad. Mcd. Sci. (India) 19 (1983) 12-21.
24. TUIIERCULOSIS AND RESPIRATORY DISEASES RE-SEARCH CENTRE, KENYA MEDICAL RESEARCHINSTITUTE. Report of Kenya National Tubercu-losis Survey 1979-1981, Phase 1: Ministry ofHealth, Kenya, 1984.
